Stock Market Non-Participation
The non-participation puzzle is the empirical fact that a large fraction of households — including many wealthy ones — hold zero direct or indirect equity, even though the historical equity premium is high enough that any household with sensible risk aversion should invest a positive amount. The puzzle is not whether everyone holds the same share of equity (life-cycle and risk-tolerance differences explain that) but whether the share is zero, which is hard to rationalize for any household with positive wealth and the standard assumption of decreasing relative risk aversion.
Magnitude. In , only of U.S. households held any stock, and even among households with more than $ in liquid assets, only did. Participation has risen substantially since (driven largely by retirement accounts), but cross-sectional and cross-country gaps remain large. Participation is positively related to education, income, and wealth — and even conditional on participation, the share of liquid wealth in equity rises with wealth, contradicting the constant-relative-risk-aversion benchmark.
Rational explanation: transaction costs. Vissing-Jorgensen (2003) shows that even small per-period participation costs (information costs, time costs, the cognitive overhead of learning to invest) can rationalize non-participation for households with low expected gains from holding equity. The model predicts that participation should rise with wealth (because the gains scale with wealth while the cost is fixed) — which matches the cross-section. The challenge for the rational explanation is to also account for the increase in participation over time and the cross-country differences, neither of which is naturally driven by transaction costs.
Behavioral explanations. Several candidates extend the rational story. Ambiguity aversion (households dislike not knowing the distribution of stock returns and abstain rather than face the unknown), loss aversion combined with narrow framing (the disutility of paper losses outweighs the expected return gain), and trust deficits (households who don’t trust financial markets stay out) all generate non-participation in environments where standard expected-utility models would predict participation.
Intuition
The basic rational benchmark says: stocks earn an equity premium of about – per year over bonds, with volatility of about . For any household with constant relative risk aversion below about , the optimal portfolio holds some equity — and at typical risk-aversion levels of –, the optimal share is or more. The fact that the median household holds zero contradicts the model not by a small amount but by the entire predicted optimum. Either risk aversion is much higher than survey data suggest, or there is a fixed cost of participating that the standard model leaves out.
The fixed-cost explanation has appeal because it generates the right cross-sectional pattern — wealthier households participate more, exactly because the cost is fixed and the benefit scales with wealth. But the cost has to be substantial to keep middle-class households out, and it is hard to identify what the cost actually is in dollars: brokerage fees are essentially zero, mutual funds are widely available, and information about index investing is free. The cost must therefore be psychological or cognitive — the discomfort of facing volatile prices, the time required to learn, or the worry of choosing wrong. Once the cost is reframed this way, the line between “rational fixed cost” and “behavioral barrier” becomes blurry.
Properties
(Equity Premium Puzzle Mirror) Non-participation is the household-side counterpart of the equity premium puzzle.
The equity premium puzzle (Mehra-Prescott 1985) asks: why is the historical equity premium so large given standard preferences? The non-participation puzzle is the dual: given the historical equity premium, why don’t all households participate? Both puzzles point at the same gap between standard preferences and observed behavior — either the premium is too high or households are leaving free money on the table. The gap is closed in some models by reframing risk aversion (Epstein-Zin), in others by introducing rare disasters, and in others by adding participation costs to the household side.
(Cross-Sectional Pattern) Participation rises with education, income, and wealth.
The Survey of Consumer Finances repeatedly finds that participation rates climb sharply with income and wealth. In the bottom quintile of wealth, participation hovers below . In the top quintile, it exceeds . Within the top decile, even the rare non-participants are an interesting puzzle — at their wealth level the foregone equity premium is hundreds of thousands of dollars annually. Participation also rises with education at any given wealth level, suggesting that information or cognitive barriers play a role beyond pure financial constraints.
(Time-Series Pattern) Participation has risen substantially since the 1980s.
Aggregate participation grew from in to over by the late s, driven primarily by the shift from defined-benefit to defined-contribution retirement plans. The (k) infrastructure forces a participation decision at the moment of employment, and default options now usually include equity exposure. The time-series pattern is hard to reconcile with a stable transaction-cost story (information costs presumably fell, but they were never the actual barrier). The leading interpretation is that institutional design — auto-enrollment, default funds, employer matching — overcame the behavioral inertia that had been keeping households out.
(Cross-Country Differences) Participation varies enormously across developed economies.
Participation rates in ranged from above in Sweden to below in Italy. The cross-country gap is too large to be explained by income or wealth differences. Trust in financial markets, the structure of the pension system, and historical familiarity with equity investing all appear to matter. Guiso, Sapienza & Zingales (2008) document a strong correlation between general trust in others and stock-market participation, suggesting that the barrier is not technical but social.
(Cost of Non-Participation) Foregone returns are large in lifetime terms.
A median U.S. household with years of working life and even modest savings forgoes substantial wealth by avoiding equity. Compounded over a working life, a equity premium translates into roughly a x wealth multiple (relative to bonds-only), which for a household saving $ over its lifetime means roughly $ in foregone retirement wealth. The non-participation puzzle is therefore not a small inefficiency at the margin but a first-order welfare loss for the households affected.
Examples
A wealthy household choosing not to participate
A household with $ in liquid wealth, no debt, and a stable income of $ a year holds its entire portfolio in CDs and money-market funds. Under the standard rational benchmark with constant relative risk aversion of and a equity premium, the household should hold roughly in equity, or $. Foregone expected return: 0.05 \cdot \150{,}000 = $7{,}50020 years of pre-retirement, the foregone wealth is roughly \$$250{,}000 in expected value. The transaction-cost explanation would have to assume that the per-period cost of holding equity exceeds $ — implausible at this wealth level. The behavioral explanation invokes ambiguity aversion (the household does not know the true distribution of returns) or trust deficits (the household does not trust financial markets) as the binding barrier.
The 1984 SCF non-participation snapshot
Survey of Consumer Finances data from — well before the rise of (k) auto-enrollment — show that only of U.S. households held any stock, directly or indirectly. Stratifying by liquid wealth: among households with less than $ in liquid assets, participation was below . Among those with $–$, participation was around . Among those with more than $, participation was . Even at the top, more than half of wealthy households were entirely out of the equity market. By , after years of (k) growth and the dot-com bull market, the comparable participation figures were roughly overall and for the top wealth tier. The shift is one of the largest behavioral changes in U.S. household finance history, and it was driven primarily by changing default options rather than by changing preferences.
See Also
- 8.10. Behavioral Household Finance — parent hub.
- 8.10.2. Under-Diversification — sibling: even households that do participate hold poorly diversified portfolios.
- 8.3. Challenges to Rational Pricing — the equity-premium puzzle that non-participation mirrors.